Journal · AI & the Future

How to get recommended by ChatGPT, not just found on Google

Guide18 Nov 20265 min readBy The Blink

Your next customer may never see a list of ten blue links. They'll ask an AI, and get three names. Here's how to be one of them.

Ask ChatGPT for "a good branding agency in Athens", "the best boutique hotel in Halkidiki" or "a reliable accountant for freelancers in Thessaloniki", and you don't get a page of results. You get a short, confident answer with a handful of names.

If you're one of those names, congratulations: you've just been recommended by the most persuasive salesperson your customer has ever met. If you're not, you don't exist in that conversation. There's no page two.

This shift has a few names: GEO (generative engine optimisation), AEO (answer engine optimisation), LLM SEO or just ChatGPT SEO. The labels will settle. The idea won't change: make your business easy for AI to find, understand and recommend.

How AI assistants choose who to recommend

It helps to know roughly what's happening under the bonnet. When someone asks an AI assistant for a recommendation, the answer comes from two places:

  1. What the model already "knows": patterns from the huge amount of text it was trained on. If your brand was talked about widely and consistently before the model's training cut-off, it may already associate you with your category.
  2. What it looks up live: assistants with web search (ChatGPT search, Perplexity, Gemini, Google's AI Overviews) pull current pages, then summarise and cite them.

You can't do much about the first one quickly. You can do a lot about the second, and over time the second feeds the first.

What both reward is the same thing: clear, consistent, credible information about you, in many places.

Seven steps to get into the answer

1. Say plainly what you do, where and for whom

AI can't recommend you for something your website only hints at. "We craft meaningful experiences" tells a model nothing. "Boutique hotel with 24 rooms in Kassandra, Halkidiki, adults-only, open May to October" tells it everything.

Put the facts in text, not just in images or video: services, locations, prices or price ranges, opening hours, who you serve. Use the words your customers use.

2. Answer the real questions people ask

Look at your enquiry inbox. The questions customers email you are the questions they're now asking AI. Answer them on your site in plain language: a proper FAQ, service pages that address objections, guides like this one. Each clear answer is something an assistant can lift and cite.

3. Make your pages machine-readable

Good structure helps people and machines alike:

4. Don't lock the door on the crawlers you want

Check your robots.txt. Some sites block AI crawlers wholesale without realising it. You may reasonably choose not to let your content train models, but search crawlers like OpenAI's OAI-SearchBot and PerplexityBot are what fetch pages for live answers. Block those and you opt out of being cited. Decide deliberately, crawler by crawler.

5. Get mentioned where AI looks

AI assistants trust brands that other people talk about. That means:

The key word is consistent. If your name, address, phone number and description differ across the web, you look less like one trustworthy business and more like three uncertain ones.

6. Earn real reviews, and answer them

Reviews are one of the strongest trust signals for both people and AI. When an assistant recommends "highly rated" places, it's reading those ratings and the words inside them. We wrote a separate guide on how to get more Google reviews without being pushy.

7. Keep doing the SEO basics

AI search didn't replace SEO. It's built on top of it. Pages that rank well, load fast and are linked to from reputable sites are far more likely to be fetched and cited. GEO is not a separate trick. It's SEO done properly, with an eye on how answers get written.

How to check where you stand

Do it by hand first. It's free and surprisingly revealing:

  1. Write down ten questions a potential customer would ask before buying what you sell.
  2. Ask them in ChatGPT, Gemini, Perplexity and Google (look at the AI Overview at the top).
  3. Note who gets recommended, what's said about them, and which sources get cited.
  4. Ask directly about your brand: "What is [your business] known for?" Then correct, on your own site and profiles, anything that comes back wrong or vague.

Repeat every month or two. The answers change as the models and the web change.

What not to do

The bottom line

Being recommended by AI comes down to the same things that have always made a business recommendable: being clear about what you do, being good at it, and having other people say so. The difference is that now a machine is listening, and repeating it to your next customer.

GEO is part of our Performance & Growth work, alongside SEO, paid media and analytics. If you want to know how your brand currently shows up in AI answers, ask us.

Written by the team at The Blink, Athens. AI-accelerated, human-edited.

Start a project
Keep reading

Got a brief?
Let's talk.

Start a project More from the Journal